Fig. 4

Changing domain borders are associated to dosage compensation features. a Example of domain borders identified by LSD classified by their concordance between male and female samples in sex-sorted embryos Hi-C data (binned at 3.5 kb, normalized by chromosome-wise ICE). A representative 1 Mb region of chrX is shown, with contact frequencies (colour bars on the side) for male (upper half) and female (lower half) data. The directionality index is shown adjacent to the Hi-C maps. Domain boundaries are marked based on their detection in male or female sample: disappearing (red triangles), appearing (green triangles) and same, i.e. unchanged (black diamonds) domain boundaries. MSL binding sites are also marked (red diamonds). The log₂ ratio of male over female insulation score is also shown in the middle track, above the domain boundary marks. b Enrichment (log₂ observed over expected ratio) of MSL binding sites around the domain borders grouped by differential pattern is shown. Three alternative definitions of MSL binding sites from three different laboratories were considered: Kuroda⁴, Becker⁵ and Akhtar⁸ laboratories, respectively. The expected frequency was estimated assuming random uniform distribution of MSL binding sites along chrX. c Enrichment (log₂ observed over expected ratio) of CLAMP binding sites around the domain borders grouped by differential binding pattern is shown. Three groups of CLAMP binding sites as defined in Soruco et al.⁴⁷ are considered: MSL-dependent binding of CLAMP (A), Partially dependent on MSL binding (B) and MSL independent binding of CLAMP (C). The expected frequency was estimated assuming random uniform distribution of CLAMP binding sites along chrX. d After classification of boundaries on the basis of combinatorial presence of BEAF-32 and CP190 or Chromator (columns), for each class the proportion of boundaries grouped by differential pattern across sexes (rows) is shown. e The log₂ ratio of insulation scores (y axis, female over male ratio) is shown for each boundary class as defined by combinations of insulators. For each box the median is marked as horizontal line, the boxes mark the interquartile range (IQR), the whiskers extend up to 1.5 IQR and individual data points are shown for outliers beyond this range.